ABSORPTION AND DIFFUSION OF HYDROGEN IN STEELS

UDK 669.14.018.85:669.778:620.193 Izvirni znanstveni ~lanek

ISSN 1580-2949 MATER. TEHNOL. 34(6)331(2000) H. J. GRABKE, E. RIECKE: ABSORPTION AND DIFFUSION OF HYDROGEN IN STEELS

ABSORPTION AND DIFFUSION OF HYDROGEN IN STEELS ABSORPCIJA IN DIFUZIJA VODIKA V JEKLIH Hans Jürgen Grabke, Ernst Riecke Max-Planck-Institut für Eisenforschung GmbH, Postfach 140444, 40074 Düsseldorf, Germany Prejem rokopisa - received: 2000-12-11; sprejem za objavo - accepted for publication: 2000-12-20 To understand and control hydrogen-induced cracking and hydrogen-induced stress-corrosion cracking in steels, the fundamental processes of hydrogen absorption and diffusion are of interest. The electrochemical hydrogen permeation method has been extensively applied to obtain data on hydrogen absorption and diffusion in iron, iron alloys and steels. Steady-state permeation measurements yield the permeation coefficient and the hydrogen concentration and activity aH, which are established on the surface in electrolytes. This information was used to study the effects of alloying elements on aH present during corrosion. Permeation transients from non-steady-state measurements yield the hydrogen diffusivity, which is affected by the presence of hydrogen traps. For shallow traps with low binding energies a distribution equilibrium between trap and lattice sites applies, deep traps are already saturated at low aH and hydrogen in deep traps is immobile at ambient temperature. The effects of Mo, V, Nb, Ti, Zr and their carbides and nitrides on the hydrogen diffusivity were studied and information was derived on the numbers and binding energies of traps which are caused by strain fields around big atoms of alloying elements or in the neighbourhood of their precipitates. For some pipeline steels with different microstructures, and after welding simulation, hydrogen diffusivities and trapping were studied. The results were related to hydrogen-induced stress-corrosion cracking, which was observed in constant-extension-rate tests. The work of fracture decreased with increasing hydrogen activity, i.e. the mobile hydrogen is decisive for failure, and the hydrogen in deep traps has no obvious effect on the fracture behaviour of steels. Key words: iron alloys, pipeline steels, hydrogen activity, permeation, diffusivity, fracture energy Za razumevanje in kontrolo pokanja zaradi vodika in napetostnega korozijskega pokanja jekel je treba poznati temeljne procese absorpcije in difuzije vodika. Metoda elektrokemi~ne permeacije vodika se je mnogo uporabljala za pridobivanje podatkov o absorpciji in difuziji vodika v èlezu, njegovih zlitinah in jeklih. Iz meritev permeacije v stacionarnem stanju so bili dolo~eni koeficient permeacije ter koncentracija in aktivnost vodika na povr{ini kovine v elektrolitih. Metoda je bila tudi uporabljena za dolo~anje vpliva legirnih elementov na aktivnost vodika med korozijo. Tranzienti permeacije pri nestacionarnih meritvah dajo podatke o difuzivnosti vodika, na katero vplivajo prisotne pasti. Za plitve pasti z majhno vezno energijo se vzpostavi ravnote`je med pastmi in mre`nimi mesti. Globoke pasti so nasi~ene è pri majhni aktivnosti vodika, vodik v njih pa je nepremi~en pri temperaturi okolice. Raziskani so bili vplivi Mo, V, Nb, Ti, Zr ter njihovih karbidov in nitridov na difuzivnost vodika in pridobljeni podatki o {tevilu in vezni energiji pasti. Te nastanejo zaradi napetostnih polj okoli velikih atomov legirnih elementov in v okolici njihovih precipitatov. Za nekatera jekla za cevovode z razli~no mikrostrukturo, tudi po simulaciji varjenja, sta bila raziskana difuzivnost in ujetje vodika v pasti, povezana z vodikom, induciranim napetostnim korozijskim pokanjem, ki je bilo opredeljeno z nateznimi preizkusi s konstantno hitrostjo iztezanja. Delo preloma se zmanj{uje z ve~anjem aktivnosti vodika, kar pove, da je mobilni vodik odgovoren za prelom, vodik v globokih pasteh pa ne vpliva na prelomno vedenje jekel. Klju~ne besede: èlezove zlitine, jekla za cevovode, vodik, aktivnost, permeabilnost, difuzivnost, energija preloma

1 INTRODUCTION Hydrogen in iron is an important and intensively studied system 1-11. Severe failure can be caused by the hydrogen embrittlement of steels, which appears as: surface blistering; hydrogen-induced cracking (HIC); stress-corrosion cracking (HISCC); classical, weld and stress cracking and hydrogen-environment embrittlement (HEE). After adsorption and dissociation hydrogen is dissolved from the atmosphere by absorption according to: H2 (gas) = 2H (dissolved)

(1)

and its concentration follows ’Sieverts law’ cH = KS ⋅ (pH2)1/2 MATERIALI IN TEHNOLOGIJE 34 (2000) 6

(2)

where cH [mol/cm3] is the concentration of dissolved H atoms, KS is Sieverts constant and pH2 [bar] the hydrogen pressure in the environment. The solubility at 1 bar H2 10-14 is extremely low at room temperature, just 3⋅10-6 at %, but it rises with temperature to 1.6⋅10-2 at% at 900°C in α-Fe. This corresponds to 3cm3 of H2 per 100g Fe. In γ-Fe the solubility is a little higher, 2.3⋅10-2 at% at 900 °C and in the melt cH is much higher, 0.13 at% at 1540 °C. As a consequence it is dangerous to saturate steels with hydrogen at elevated temperatures, without the possibility for H2-desorption during cooling. This desorption may be hindered by oxide scales, enamel or zinc coatings; and an oversaturated solution may result in a variety of failures: shatter cracks, flakes, fish eyes, etc. 331

H. J. GRABKE, E. RIECKE: ABSORPTION AND DIFFUSION OF HYDROGEN IN STEELS

Hydrogen absorption occurs not only from gaseous hydrogen, but more often from electrolytes, e.g. during pickling or the corrosion of steels. The most important reactions during the corrosion of iron in an acid environment are 15-18: the anodic dissolution of Fe (Volmer reaction): Fe + 2 H+ = Fe2+ + 2 Had

(3a)

the recombination of adsorbed H (Tafel reaction): 2 Had = H2 (gas)

(3b)

and the absorption of Had: Had = H(dissolved)

(3c)

The interplay of these reactions establishes a hydrogen activity aH on the metal surface which is strongly affected by inhibitors and promoters in the electrolyte and by the presence of alloying elements and impurities in the metal surface 19-25. The study of these latter effects was the subject of our studies 26-28. The hydrogen activity is defined as the hydrogen concentration cH related to the hydrogen concentration c 0H at T = 0°C and at 1 bar H2 aH =

cH c OH

 pH  =  O2   pH2 

1 /2

(4)

If the anodic dissolution reaction (3a) is enhanced (e.g. by the presence of sulfur) and/or the recombination reaction (3b) is retarded (e.g. by arsenate) very high hydrogen activities can occur on steel surfaces which cause the severe absorption of hydrogen. Hydrogen is very dangerous, especially in ferritic steels, since it generally diffuses very quickly to defects, where it exerts its deleterious effects. In fact, the diffusivity of H in pure and undeformed α-iron at room

Figure 1: Arrhenius plot of hydrogen diffusion coefficient (DH) in pure annealed iron, own data 6, 7 and nearby Arrhenius lines from other authors 29-32 Slika 1: Arrheniusov diagram odvisnosti koeficienta difuzije DH v ~istem àrjenem èlezu, lastni podatki 6,7 in pribli`ne Arrheniusove ~rte za druge avtorje 29-52

332

temperature is very high, about 1⋅10-4 cm2/sec 27-32, this compares with about 1⋅10-16 cm2/sec for carbon and nitrogen. Our own careful studies on very pure iron yielded 6,7: DH = (5.12 ± 0.6) ⋅ 10-4 exp (-(4.15 ± 0.3)kJ mol-1/RT) (5) The purer and better recrystallized the iron samples the higher were the data obtained, see Figure 1. The activation energy is strikingly low: in comparison with about 80 kJ/mol for carbon and nitrogen. This indicates that hydrogen moves through the ferritic lattice as a proton, tunneling through the interstitial sites. Neither grain boundary nor surface diffusion are faster processes. Alloying elements and deformation generally decrease the diffusivity of H in iron and a tremendous number of Arrhenius lines have been published over the years, on hydrogen diffusion in various steels, see Figure 2 6,7,33. The data in Figure 2 were derived using the assumption that hydrogen concentration and diffusion follow Fick’s laws, and for the higher temperature range the data obtained could be approximately extrapolated to equ. (5), whereas the data at lower temperatures are mostly on lower Arrhenius lines and would yield higher activation energies. This diffusion behaviour of hydrogen was treated by McNabb and Foster 33 by applying a trapping model: ie. the hydrogen atoms

Figure 2: Arrhenius plot of hydrogen diffusivities (Deff) measured by numerous authors in different steels, compared to DH in pure recrystallized iron, and the H diffusivity in 60% cold-rolled iron which can be described by equation (12) with Nft = 6.7⋅1019/cm3Fe and ∆H°ft = -36 kJ/molH 6, 7 Slika 2: Arrheniusov diagram odvisnosti koeficienta difuzije DH po meritvah mnogih avtorjev v razli~nih jeklih v primerjavi z DH v ~istem rekristaliziranem èlezu in difuzivnost H v 60 % hladno valjanem jeklu, ki jo opisuje ena~ba (12) z Nft = 6,7⋅1019/cm3Fe in ∆H°ft = -36 kJ/molH 6, 7 MATERIALI IN TEHNOLOGIJE 34 (2000) 6


wander in a random manner through the lattice, but tend to get trapped or delayed at sites which act as ’traps’. These sites are regarded as potential wells of greater depth, than those in the undisturbed, defect-free crystal lattice. During the hydrogen diffusion, a local equilibrium is assumed for the ’trapping reaction’: H(in normal, interstitial site) + trap = = H(in trap) + free normal, interstitial site

(6)

This situation corresponds to an adsorption equilibrium where: rate of adsorption = rate of desorption: k⋅cH (1-θ)⋅Nt = k'⋅ct⋅N

(7)

where Nt is the number of traps and θ is their degree of occupation, N is the number of normal interstitial sites and ct = θ⋅Nt is the concentration of H in the traps. The quotient of the rate constants for the forward and backward reactions is the equilibrium constant of the trapping reaction (6): k θ⋅ N = = K = exp (-∆Gt0 / RT) k' c H (1 − θ) = exp (+∆St0 / R) exp (-∆Ht0 / RT) and for the degree of trap occupation: c ⋅K / N β β = cH ⋅ K/N θt = H = 1+ cH K / N 1+ β

(8) (9)

In steels there are different kinds of traps, in principle a more-or-less continuous spectrum of ’binding energies’ ∆Ht° and values of K is to be expected, however, we usually distinguish between flat traps with -∆Hft < 30 kJ/molH and β 50 kJ/molH and β >> 1 The deep traps are already saturated at low values of aH and cH and do not affect the diffusivity DH, whereas the occupation of the flat traps varies with the increase and the decrease of aH resp. cH in diffusion experiments or during corrosion. Non-steady-state diffusion experiments are generally described by Fick’s second law, which in the presence of traps describes the H concentration changes in the normal interstitial sites and in the flat traps:  ∂ 2 cH  ∂c H ∂c ft + = DH   ∂t ∂t  ∂x 2 

(10)

this equation can be rewritten:  ∂ 2 cH  ∂c H DH =   1 + ( ∂c ft / ∂c H )  ∂x 2  ∂t

(11)

A comparison with the usual formulation of Fick’s second law shows that in the presence of flat traps an effective diffusivity of H is observed, which is given by DH D (12) D eff = = H 1 + ( ∂c ft / ∂c H ) 1 + α MATERIALI IN TEHNOLOGIJE 34 (2000) 6

Since cft = θ ⋅ Nft we obtain α = Kft ⋅ Nft / N = (Nft / N) exp (∆Sft0 / R) exp (-∆Hft0 / RT)

(13)

Thus, hydrogen diffusivity decreases with a high occupancy of flat traps Nft and high binding energies I∆H°I, and the Arrhenius plots, as shown in Figure 2, can be explained by the trapping model and equation (12). Figure 2 also demonstrates that the numerous studies on different steels might well be explained by the effects of strained regions and dislocations 6,7. The hydrogen diffusivity in iron which has been 60% deformed by cold-rolling corresponds to the diffusion behaviour in many different steels with various degrees of deformation. It is probable that the flat traps correspond to dilated sites near dislocations and the deep traps to dislocation cores. Such dilated sites are also caused by large atoms of alloying elements and by the stresses around precipitates. Another aim of this study was to clarify the effects of alloying elements such as Ti, Zr, V, Nb, Cr and Mo and their carbides and nitrides on hydrogen solubility and diffusivity, to obtain a better understanding of traps. Studies were conducted on H permeation, diffusion and solubility in several binary alloys of iron and the above mentioned elements, see table 1 34-40. 2 EXPERIMENTAL 2.1 Electrochemical Double Cell Hydrogen absorption, permeation and diffusion were measured using the electrochemical double cell, first described by Devanathan and Stachurski 41,42. An iron or steel membrane is charged with hydrogen from the entry side, at the exit side the hydrogen is oxidized anodically and the anodic current Ip is a measure of the hydrogen permeation. The schematics in Figure 3a demonstrate the principle of the method. Two techniques for hydrogen charging have been applied a) electrolytic charging from an aqueous solution or b) gas-phase charging from pure H2 or N2-H2 mixtures. The application of both techniques us allows to measure at low and high hydrogen activities and at a very well defined standard activity aH = 1, corresponding to pH2 = 1 bar. For the study of the compositional or structural effects of H absorption, the entry side remained uncoated; this was also the case for sufficiently high cathodic polarisation currents, obtained at constant cathodic current densities iC or cathodic potentials EC. Otherwise, especially in the case of hydrogen absorption from the gas phase, the entry side of the membrane had to be Pd coated, to avoid the inhibition of hydrogen adsorption and absorption by oxide films. The exit side of the membrane was always coated with Pd, and polarised anodically at Ea = 200 mV(SHE) in a 0,1N 333


NaOH solution, deaerated by bubbling with N2. Here the hydrogen is oxidized immediately and its concentration is virtually zero. The anodic permeation current (Ip) is corrected by a blind current, measured when no hydrogen was charged.

Thus, the hydrogen activity can be obtained if the permeation coefficient Φ° is known from the experiment described above, and Φ° must be determined separately for each material, since its value is dependent on the steel composition and microstructure.

2.2 Determination of the permeation coefficient

2.4 Determination of Hydrogen Diffusivity

The permeation coefficient is the product of the hydrogen solubility at 1 bar H2 and the hydrogen diffusivity in a steady-state experiment:

Upon charging or discharging the membrane in the electrochemical double cell, current transients are observed (see Figure 3b). These transients correspond to the absorption of H into the material during charging and the emptying by anodic oxidation during discharging. The process does not only involve the hydrogen in ideal solution, i.e. in the normal interstitial lattice sites, but also the flat traps are filled and emptied, whereas the deep traps are saturated in the very first charging step (see Figure 3b) and H sticks to these traps during

Φ 0 = c 0H ⋅ D H

(14)

Its value was determined for each material investigated by charging the Pd-coated membrane from the atmosphere, i.e. pure hydrogen at 1 bar. After attaining a steady state, the flow of hydrogen (jH) is given by Fick’s first law: jH = D H

c 0H − c eH = D H ⋅ c 0H / d = Φ 0 / d d

(15)

where d is the membrane thickness; the hydrogen concentration at the entry side (cH°) and at the exit side (cHe) = 0. The anodic oxidation of the flux of hydrogen arriving at the exit side yields the permeation current: Ip = jH ⋅ F ⋅ A

(16)

where F is the Faraday constant and A is the area of the membrane. The permeability results from: Φ0 =

I p ⋅ d  mol  = c 0H ⋅ D H   F ⋅ A  cmsec 

(17)

2.3 Measurement of hydrogen activity In electrolytes, hydrogen activitiy on iron and steel surfaces is the result of an interplay of electrochemical reactions (3). The value of aH can be very high due to the promotion of reaction (3a), the discharge and H adsorption, or to the inhibition of reaction (3b) - the H recombination (Tafel) reaction. It is of great scientific and technical interest to know about effects of the electrolyte and of the steel composition on the established hydrogen activity, e.g. during pickling or corrosion. In this study, permeation measurements were used to obtain the hydrogen activity (aH) during corrosion in 1 M H2SO4. The hydrogen concentration (cH) at the entry side of the corroding membrane is determined using: cH =

I p ⋅d

(18)

F ⋅ A ⋅ c 0H ⋅ D H

and related to the hydrogen concentration cH° at aH = 1 (pH2 = 1 bar) by: aH = 334

cH c

0 H

=

I p ⋅d F ⋅ A ⋅ c ⋅D H 0 H

=

I p ⋅d F ⋅ A ⋅Φ 0

(19)

Figure 3: Schematics of measurements in the electrochemical double cell: a) schematic drawing of the double cell, hydrogen charging from the left side, from electrolyte or gas phase, anodic oxidation of the permeating hydrogen in the right cell - measurement of the permeation current IP; b) schematic presentation of a permeation experiment, in the first step saturation of the deep traps at low charging current iC and low hydrogen activity aH, then transients of hydrogen permeation for determination of the effective diffusivity Deff Slika 3: Shema meritev v dvojni elektrokemi~ni celici: a) Shema dvojne celice, nasi~enje z vodikom z leve strani iz elektrolita ali iz plina, anodna oksidacija permeatiranega vodika v desni celici meritev toka permeacije Ip; b) Shema poskusa permeacije, v prvi stopnji nasi~enje globokih pasti pri majhnem toku ic in majhni aktivnosti vodika aH, nato tranzienti permeacije vodika za dolo~itev dejanske difuzivnosti Deff MATERIALI IN TEHNOLOGIJE 34 (2000) 6


long-term discharging. In our experiments for determining the hydrogen diffusivity the deep traps were always saturated in the first charging step at a low charging current (iC) and hydrogen activity (aH) and then the transients of charging and discharging were measured. These non-steady-state experiments lead to a DH for pure and undeformed iron, and according to equation (12), to a Deff, which is affected by the flat traps. Such a measurement is demonstrated in Figure 3b, where the permeation current during a sequence of galvanostatically controlled charging steps is plotted schematically. In the first step at (iC)O a low hydrogen activity is established on and in the material and all the deep traps are saturated. Then the cathodic current is raised and lowered at a higher level, in the range of iC and aH, where the flat traps can be occupied and released. The permeation transients obtained in that range are described by a series I p (t) − I Sp I p (0) − I Sp

∞

= 2∑ ( −1) n −1 exp( − n 2 π 2 τ)

(20)

n =1

where τ = Deff · t / d2. For the determination of Deff from the breakthrough time (tb) (see Figure 3b), the first three terms of the right hand side of equation (20) are used to give: Deff ≈ d2 / 20 · tb

(21)

The breakthrough time is defined by the intersection point of the time axis with the tangent drawn at the inflection point of the permeation curve (see Figure 3b). In the first charging of a hydrogen-free sample, all the flat and deep traps are effective and the diffusivity is not defined. The trapping effect of a steel is characterized by the retention of the permeation through the membrane, i.e. the ’time-lag’ till the hydrogen appears at the exit side after the start of charging. A measure for the amount of hydrogen trapped is the charge QT corresponding to the hatched area in Figure 3b, which can be obtained by the integration of the permeation curve Ip(t) to te. If the hydrogen concentration at the entry side is kept constant, the trapping effect is described by: tT = 2Q / Ip

(22)

For pure trap-free iron tL = 2Q / Ip = d2 / 6 DH

(23)

where DH is the hydrogen diffusivity in iron without a significant trapping effect. The theory of McNabb and Foster leads to an expression describing the retardation of hydrogen diffusion by the action of flat traps using equations (12) and (13). Thus from the non-steady-state permeation curves the number of flat traps (Nft) can be determined. The binding energy of the hydrogen atoms in the flat traps and their concentration can be obtained using equation MATERIALI IN TEHNOLOGIJE 34 (2000) 6

(12) for calculating Deff from measurements at different temperatures. (This evaluation leads to a value for ∆Hft0, and ∆Sft0 is assumed to be small). From time-lag measurements the effect of deep traps can be derived using the theory of McNabb and Foster 33 which leads to the expression:   2 2 − 2 ⋅ ln(1 + β dt )  + α ft 1 +  β dt β dt  = N dt ⋅ K dt / N β dt = c H ⋅ K dt / N

tT  3α dt − 1 =  β dt tL  with α dt

(24)

Measurements of the time lag as a function of the hydrogen pressure can be fitted according to equation (24) and yield the number of deep traps and the binding energy, in this case ∆Gdt0 is obtained. The value of αft for the flat traps must be known from non-steady-state measurements of the permeation transients in a range of elevated aH 40. 3 RESULTS 3.1 Effects of C, S, P, Mn, Si, Cr, Ni, Sn and Cu on corrosion and H absorption During the pickling of steels hydrogen is often absorbed which may cause defects and failures in the material. The hydrogen absorption is affected by components in the electrolyte, but also by the composition of the steel surface on which alloying elements and impurities may be enriched after hot rolling or after some dissolution in the pickling acid. The retarding effects of Cu on corrosion and H-absorption have been reported 21,22. Phosphorous is reported to enhance the corrosion and H absorption, while H absorption 19,20,21 is also increased by S, Se, As, Sb and Sn 19,23,24,25. These effects have been investigated quantitatively for various binary Fe alloys (see table 1), using the electrochemical double cell 26,27. Table 1: Compositions (wt%) of investigated alloys (only main alloying addition, otherwise relatively pure, see original papers) Tabela 1: Sestava (mas.%) raziskanih zlitin (samo glavni legirni elementi, sicer ~iste zlitine, glej izvirne ~lanke) Binary alloys/Binarne zlitine 26-28 Fe-0.20 C, Fe-0.046%S, Fe-0.052%P, Fe-0.1%P, Fe-3.08%Si, Fe-1.53%Mn, Fe-5.19Cr, Fe-10.4%Cr Fe-5.07%Ni, Fe-10.2%Ni, Fe-0.55%Cu, Fe-3.4% Cu, Fe-0.17%Sn Fe-0.21%Ti, Fe-0.19%V, Fe-0.3%Zr, Fe-0.35%Nb, Fe-0.36%Mo Ternary alloys/Ternarne zlitine 34-40

alloys Fe-Me-C Fe-Ti-C Fe-V-C Fe-Zr-C Fe-Nb-C Fe-Mo-C

% Me

%C

0.22 0.19 0.27 0.35 0.33

0.074 0.081 0.067 0.066 0.065

alloys Fe-Me-N Fe-Ti-N Fe-V-N Fe-Zr-N Fe-Nb-N Fe-Mo-N

% Me

%N

0.18 0.20 0.63 0.35 0.37

0.042 0.048 0.081 0.043 0.061

For the determination of the hydrogen activities and concentrations established during corrosion the 335


Figure 4: Hydrogen permeation and diffusion in binary Fe alloys 26: a) permeation coefficient Φ° measured at pH2 = 1 bar, b) hydrogen diffusivities Deff measured after saturation of the deep traps Slika 4: Permeacije in difuzija vodika v binarnih zlitinah èleza 26: a) koeficient permeacije Φ°, izmerjen pri pH2 = 1 bar, b) difuzivnost vodika Deff, izmerjena po nasi~enju globokih pasti

elements, Ni is not clearly enriched, but minor dissolution of Sn and Cu leads to the enrichment of these elements. Current density potential curves measured on these binary alloys in 1M H2SO4 indicate, by their shift after some time of corrosion, if anodic or cathodic reactions are changed by these enrichments 28, see Figure 5. The most interesting data are the mass loss by corrosion, and the hydrogen activity established during corrosion, which have been recorded after periods of corrosion in 1M H2SO4 at 25 °C, see Figure 5. The data show a strong increase in corrosion by S and P, whereas Ni, Sn and Cu reduce the amount of corrosion, the other elements have minor effects. The high mass loss of the Fe-Mn alloy was obviously due to its relatively high S content (80 wt.ppm compared to 20 wt.ppm in the unalloyed Fe). The effects of S and P are most probably caused by the catalytic effects of sulphide and phoshide anions on the anodic dissolution of iron. Detailed studies have been conducted on the effects of Cu 18 and of P 34. The total hydrogen content of the samples after different periods of corrosion in 1 M H2SO4 was determined by chemical analysis (hot extraction). The H content of the pure Fe and the Fe-C, Fe-Ni and Fe-Cu alloys are low, around the detection limit of the method (0.4·10-6 mol H/cm3). The hydrogen content is enhanced in the Fe-Cr alloys due to the trapping effect. For some alloys the H content during corrosion in 1 M H2SO4 goes through a maximum and then decreases, e. g. for Fe-Sn caused by increasing Sn enrichment, and for Fe-P by the formation of a phosphate film. Most striking is the high H content of the Fe-S alloy. At high values of aH the amount of hydrogen in Fe and its alloys increases more than the hydrogen activity,

permeation coefficient Φ° had to be measured. Figure 4a shows that most of the alloying elements cause only a slight reduction in Φ°: the greatest effect was observed for 3,1 wt%Si and 10,4%Cr. For these alloys the H diffusivities were also determined, see Figure 4b. Most elements have only a small effect on the diffusivity, but Cr clearly retards H diffusion, indicating trapping by Cr atoms. A calculation of the hydrogen solubility is possible from cH = Φ0 / Deff, and the data for the Fe-Si alloy render a clearly decreased solubility 26. The studied alloying and impurity elements affect H permeation only weakly, Si and Cr decrease the permeation flow, Cr traps H and thus decreases DH, and Si tends to decrease the H solubility. The investigated iron alloys, see table 1, were corroded in 1M H2SO4 and analyzed afterwards by taking AES sputter profiles 27. Carbon and sulfur are enriched in the surface after corrosion, as is phosphorus, which according to XPS spectra is present as both phosphide and phosphate. In contrast, silicon and chromium are depleted near the surface, these elements are obviously dissolved preferentially. Of the nobler

Figure 5: Results of corrosion experiments in 1 M H2SO4 at 25 °C on binary Fe alloys 27, mass loss (∆m) and hydrogen activity (aH) after 25 hours corrosion, corrosion potential and amount of dissolved alloying element (related to the initial composition, not determined for Fe-0.2C and Fe-0.046S) Slika 5: Rezultati korozijskih poskusov na binarnih Fe-zlitinah v 1 M H2SO4 pri 25 °C 27, izguba mase ∆m in aktivnost vodika po 25 urah korozije, korozijski potencial in koli~ina raztopljenih legirnih elementov (glede na za~etno sestavo, ni dolo~ena za Fe-0,2 C in Fe-0,046 S).

336

MATERIALI IN TEHNOLOGIJE 34 (2000) 6

H. J. GRABKE, E. RIECKE: ABSORPTION AND DIFFUSION OF HYDROGEN IN STEELS Table 2: Data on permeation, diffusivity, solubility and trapping of H in ternary alloys Tabela 2: Podatki o permeaciji, difuzivnosti, topnosti in ujetju H v ternarnih zlitinah

Alloy

Φ° ⋅ 1013 mol/cm ⋅ s Fe-Ti-C 1.28 Fe-Ti-N 2.2 Fe-V-C 2.12 Fe-V-N 2.15 Fe-Zr-C 2.05 Fe-Zr-N 2.09 Fe-Nb-C 2.16 Fe-Nb-N 2.23 Fe-Mo-C 2.0 Fe-Mo-N 1.68 Fe(60% def,) 2.51

Deff ⋅ 106 cm2/s 0.72 6.1 3.5 2.7 24 3.0 0.82 5.3 42 0.27 96

cHeff ⋅ 108 mol/cm3 3.5 4.1 7.2 9.2 1.1 8.5 30 4.7 0.6 0.92 26

cHtotal ⋅ 106 mol/cm3 10 1.9 1.8 3.0 0.4 16 13 3.5 0.05 0.04

due to the formation of dislocations, defects and microcracks.

Nft ⋅ 103 mol/cm3 28 3.0 22 14 0.83 7.3 60 9.2 4.0 0.9 0.72

-∆Hft kJ/mol 20.6 20.7 17.2 18.9 19.9 20.4 18.3 18.2 13.9 19.3 27.9

Ndt ⋅ 106 mol/cm3 10 1.9 1.8 3.1 0.4 17 13 3.7 0.045 0.035 0.62

-∆Gdt kJ/mol 58.5 60.5 57.0 56.0 58.5 56.1 56.0 54.9 56.5 56.0 55.5

3.2 Effects of Mo, V, Nb, Ti, Zr and their carbides and nitrides on H diffusivity and solubility

various different traps generated by the presence of the elements and their precipitates. For the binary Fe-Me alloys the steady-state permeation measurements produced data for Φ° which are a little lower than for pure iron, the temperature

Fine-grained microalloyed steels have found many applications because of their good mechanical properties, high strength and good ductility, but they are susceptible to hydrogen embrittlement. Their microstructure is stabilized by finely divided carbides, nitrides and carbonitrides of the elements V, Nb and Ti, which pin grain boundaries and dislocations. These precipitates generate a lot of extended sites which are suitable for hydrogen atoms, i.e. hydrogen traps. Such traps affect the solubility and diffusivity of hydrogen. To study the effect of the alloying elements Me used in microalloyed steels and their carbides, binary Fe-Me and ternary Fe-Me-C and Fe-Me-N melts were prepared with about 0,2 at.% Me = Mo, V, Nb, Ti, Zr, see table 1. The ingots were forged and solution annealed at high temperature. The binary alloys were heat treated at 750 °C for 30 min, the ternary alloys for 300 h at 600 °C. The materials were characterized by optical and electron microscopy 35-40. In the Fe-Mo-C alloy, besides the coarse primary Fe3C-Mo7C3 carbides, there were fine Mo2C needles in the bulk which cause strains and dislocations in their neighbourhood due to their hexagonal structure and incoherence with the iron lattice 34-36. In contrast, the fine VC particles in Fe-V-C alloy x are finely dispersed, they are cubic and semicoherent with the Fe lattice and cause less strains. The cubic TiC, NbC and ZrC particles were less numerous and relatively coarse, due to their low solubility. The nitrides are even less soluble and during the solution annealing much less dissolution of the primary nitrides takes place, so that there are less finely dispersed nitride precipitates than carbide precipitates formed after the heat treatment 40. For these alloys it was most interesting to see, how the alloying elements and their carbides and nitrides affect the diffusivity and solubility of hydrogen by the

Figure 6: Hydrogen permeation and diffusion in ternary Fe alloys with precipitates of Ti-, V- and Nb carbides and nitrides 40: a) permeation coefficient Φ° from steady-state experiments, b) hydrogen diffusivities from non-steady-state experiments Slika 6: Permeacija in difuzija vodika v ternarnih Fe-zlitinah z izlo~ki Ti-, V- in Nb- karbidov in nitridov 40: a) koeficient permeacije Φ° za poskuse v stacionarnem stanju, b) difuzivnost vodika iz poskusov v nestacionarnem stanju


337

H. J. GRABKE, E. RIECKE: ABSORPTION AND DIFFUSION OF HYDROGEN IN STEELS Table 3: Compositions (wt%) of investigated steels Tabela 3: Sestava (mas.%) raziskanih jekel 46

46

Steel C Si Mn P S Mo St 0 0.05 0.45 1.45 0.009 0.001 0.007 St 1 - 4 similar, different thermomechanical treatments (see table 4) St 5 0.6 0.28 0.83 0.008 >.001 St 6 0.04 0.29 1.36 0.009 0.14 St 7 0.03 0.3 1.28 0.006 A516 0.188 0.267 1.05 0.008 0.001 0.055 E355 0.116 0.275 1.34 0.008 0.0005 0.059 E500 0.069 0.248 1.36 0.0009 0.0001 0.208

Ni 0.23 0.2 0.2 0.208 0.505

Cr 0.03

0.189 0.1 0.147

Cu 0.3

0.113 0.12

V 0.035

Nb 0.047

0.04 -

0.033 0.034 0.038 0.026 -

0.046

Table 4: Data on permeation. diffusion. solubility and trapping of H in low-alloy steels Tabela 4: Podatki o permeaciji, difuziji, topnosti in ujetju H v malo legiranih jeklih

Steel St 0 St 1 St 2 St 3 St 4 St 5 St 6 St 7 A 516 E 355 E 500 Fe (60

Φ° ⋅ 1013 Deff ⋅ 106 cm2/s mol/cm2⋅s 1.63 10.4 TM(γ) surface centre 1.63 16 1.65 21.5 TM(γ) 1.58 18.1 TM(α+γ) 1.58 17.6 TM(γ) + ACC Q+T 1.79 10.2 1.74 39.4 TM(γ) 1.56 26.4 TM(γ) + ACC 1.81 33.2 TM(α + γ) + ACC normalized 1.51 25.1 " 1.5 23.5 Q+T 1.6 14.5 % def.) 2.51 1.4

ceff ⋅ 108 ctotal ⋅ 108 mol/cm3 mol/cm3 2.34 9.58 1.53 8.03 1.17 4.22 1.4 5.9 1.42 6.22 2.44 27.3 0.63 6.54 0.94 4.76 0.75 6.1 0.99 8.2 1.01 7.32 1.71 48.6 17.4 75.8

dependency corresponded to the value ∆H = 34,3 kJ/molH for pure iron. Also for the ternary Fe-Me-C and Fe-Me-N alloys with carbides and nitrides, Φ° was decreased only by a tortuosity factor see table 2 and Figure 6a. This decrease is most probably caused by the reduction in the cross-sectional area free for diffusion, due to the precipitates, and possibly, regions with compressive stresses. The effect on Φ° is low for most of the binary and ternary alloys, the largest effect is for the Fe-Ti-C alloy where a high concentration of particles with sizes 1-5 nm was observed by TEM. In contrast, the non-steady-state measurements of Deff showed strong effects of the alloying elements as well as of the precipitates. H diffusivity in the Fe-Me alloys decreases strongly with increasing atomic size of the elements in the sequence V, Mo, Ti, Nb and Zr 37. While the effective diffusivity decreases, the hydrogen solubility increases, both caused by flat traps and extended interstitial sites near the Me substituting atoms. In addition, attractive chemical interactions may occur between these hydride-forming elements and the dissolved H-atoms. The binding energies are between -5 kJ/mol H for V and -21 kJ/mol H for Zr 37. The effective diffusivity is also decreased in the presence of the carbides and nitrides by about one order 338

Nft ⋅ 103 mol/cm3 0.63 0.86 1.4 2.6 3.1 0.46 8.1 4.1 7.5 3.6 1.0 5.0 0.72

-∆Hft kJ/mol 23.0 21.0 19.0 18.0 17.6 23.9 12.5 15.6 13.3 16.1 19.6 17.0 27.9

Ndt ⋅ 108 mol/cm3 7.4 7.5 3.2 5.0 6.0 29.0 6.5 5.0 6.0 7.5 6.5 5.5 62

-∆Gdt kJ/mol 58.0 53.2 56.0 54.0 52.0 53.0 54.3 51.5 53.8 56.5 57.3 53.2 55.5

of magnitude, see Figure 6b, and this effect is explained by flat traps with binding energies in the range -17 kJ/mol H (Fe-V-C) and -21 kJ/mol H (Fe-Ti-C, Fe-Ti-N). The amount of dissolved hydrogen is mainly determined by the presence of deep traps with binding energies between -55 kJ/mol H (Fe-Nb-N) and -60.5 kJ/mol H (Fe-Ti-N). The number of flat traps is much higher 10-2 - 10-3 mol/cm3 compared with 10-6 - 10-5 mol/cm3 of deep traps. However, analysis of the data shows that most of the hydrogen is in the deep traps (> 96%), the concentration in the flat traps is much smaller (1-4 %), and the ideally dissolved hydrogen is much less (< 0,1 %). This is valid at aH = 1, however, the amount of mobile H in ideal solution and in flat traps increases in proportion to aH and becomes greater than the amount of H in deep traps at the critical high values of aH > 100. Upon corrosion in 1 M H2SO4 at 25 °C the alloying elements from the binary Fe-Me alloys are dissolved without enrichment at the surface, as observed with AES. Mo and Zr cause small reductions of cathodic hydrogen formation and corrosion current. The carbides show some corrosion resistance and appear to be enriched in a surface oxide film and also have minor effects on the corrosion behaviour. Mass loss and hydrogen activity established at the corrosion potential MATERIALI IN TEHNOLOGIJE 34 (2000) 6


do not differ significantly from the values for pure iron, however, the amount of hydrogen dissolved is considerably enhanced by the alloying elements: in the sequence Mo < V < Nb < Ti < Zr and also for the Fe-V-C alloy with the finely distributed vanadium carbides and the large number of traps. 3.3 Effects of microstructure in pipeline steels Various low-alloy steels, such as those used for pipelines, were investigated in terms of hydrogen permeation, solubility and diffusion 46, they had different contents of microalloying additions, V and Nb, and were processed differently see tables 3 and 4. In the first series of experiments all the investigated steels were charged with hydrogen from the gas phase at pH2 = 1 bar and 25 °C, the permeation current was followed until it reached the steady-state permeation current Ip, this measurement yields the amount Q of H absorbed in the steel and the permeation coefficient Φ°, which was determined for the temperature range 10-80 °C. The data for the commercial pipeline steel St0 show no clear differences for samples taken from the centre and the more deformed surface region, also heat treatment at 240 °C for 80 h had no notable effect on Φ°. Furthermore, the effect of different thermomechanical treatments on Φ° is small, see table 4. In all cases, the decrease of Φ° compared to pure iron can be largely explained by the effect of the Si content. Silicon decreases the H solubility and thus causes a decrease in Φ° (see chapter 3.1). The effects of microalloying elements on Φ° decrease in the sequence V > Ti > Nb, they are caused by the precipitation of carbides. The small VCx particles are very finely dispersed, VCx and TiC are precipitates coherent with the iron matrix, both cause internal stresses and these effect a decrease of Φ°. Internal stresses also result from fast cooling after welding, thus welding simulation also causes a decrease in Φ°. Since the temperature dependence of Φ° in all cases can be described by the activation energy valid for pure iron 34,3 kJ/molH, the effects described are just decreasing the temperature-independent factor, and can be explained by the reduced cross-section of permeation, as affected by the carbides and by compressive stresses. The diffusivity Deff, as affected by the flat traps, shows much stronger deviations from the ideal value for pure iron than the permeation coefficient see table 4. Deff is decreased by increasing trap density, especially in the range of higher temperatures, and a higher binding energy causes a steeper decrease mainly in the range of lower temperatures. The effect of high binding energies is observed to correlate with high dislocation densities and internal stresses. This is true for the steels investigated in this study, in the case of the strongly deformed near-surface region of the pipeline steel St 0 MATERIALI IN TEHNOLOGIJE 34 (2000) 6

and after different thermomechanical treatments, and for the heat-affected zones (HAZ), especially after very rapid cooling from high temperatures. By tempering, the quenching effect can be partially removed, but only if the microstructure is not stabilized by fine carbide dispersions. Especially the finely dispersed VCx precipitates, which are able to stabilize a dislocation network and cause a relatively low value of Deff, more-or-less independent of heat treatments and welding simulations. The concentration of mobile hydrogen, which is effective in diffusion and steady-state permeation is high for the steels with high dislocation densities and internal stresses. This is demonstrated for the samples from the surface region of the pipeline steel, for the quenched and tempered steels (St4, E500) and for the various HAZ structures after welding simulation. The disordered microstructures are stabilized again, more or less, by the carbide formers Ti, Nb and especially V (HAZ structures). Tempering decreases the internal stresses and simultaneously the trapping effect and hydrogen concentration cHeff are reduced. Only the presence of V stabilizes these data. Information on the deep traps was obtained from time-lag measurements according to equation (24), from fitting plots of (tT - tL)/tL vs. (pH2)-1/2. The trapping by deep traps is enhanced for steels which were tempered after a quench, i.e. quenched and tempered steels and HAZ structures. For the base steel without Ti, Nb and V this effect is not observed, and we may conclude that the number of deep traps is connected to the presence of carbide precipitates. Obviously, the number of deep traps increases with the increasing number of carbide/ferrite interfaces and for incoherent precipitates. Titanium causes large effects with deep traps, due to the incoherence of TiC. Besides the effects of carbides, deformation also leads to an increased number of deep traps. The effect of deep traps depends on their number and on the binding energy. The latter influence is minor, since even at the hydrogen activity aH = 1, the saturation of traps with ∆G = -51 kJ/mol is already 73%, at aH = 10 it is 96 %. The values found for the investigated steels are between -51 and -58 kJ/mol H, and thus very near to the value determined for pure iron with a high deformation and dislocation density: -55.5 kJ/molH. In the literature, values of -59 to -60 kJ/molH are given for deep traps in dislocations and grain boundaries 8. Therefore, we can conclude that deep traps at, or in the neighbourhood of, carbides are just extended interstitial sites, as in dislocation cores and grain boundaries. Special chemical interactions of H with carbides and also with nitrides are improbable, and high binding energies of -90 kJ/mol H to TiC 43,44 are not confirmed by our work. The total hydrogen content is the sum of the hydrogen in deep traps and the mobile hydrogen. The 339


Figure 8: Relative fracture energies of some pipeline steels from constant-extension-rate tests (CERT) 46 at µ = 10-5/s: a) CERT in electrolyte at aH ≈ 500, b) CERT in electrolyte at aH ≈ 1000 - 1500, c) CERT in air after 96 h precharging and 24 h hydrogen desorption at room temperature Slika 8: Relativna energija preloma nekaterih jekel za cevovode, dolo~ena s poskusi s konstantno hitrostjo iztezanja (CERT) 46 pri µ = 10-5/s: a) CERT v elektrolitu pri aH ≈ 500, b) CERT v elektrolitu pri aH ≈ 1000-1500, c) CERT v elektrolitu po 96 urah katodnega nasi~enja in 24 urne desorpcije vodika pri sobni temperaturi

Figure 7: Results of constant-extension-rate tests (CERT) with several pipeline steels 46 in electrolyte 1 M H2SO4 + 10-3 M As2O3 under cathodic charging with hydrogen; relative fracture energy dependence on a) the extension rate µ at aH = 300, b) the hydrogen activity aH Slika 7: Rezultati poskusov s konstantno hitrostjo iztezanja (CERT) na ve~ jeklih za cevovode 46 v elektrolitu 1 M H2SO4 + 10-3 As2O3 pri katodnem nasi~enju z vodikom; relativna energija preloma v odvisnosti od: a) hitrosti iztezanja µ pri aH = 300, b) aktivnosti vodika aH

latter is in ideal solution and in flat traps, and its concentration is approximately proportional to aH. In pure recrystallized iron and steels with low strength, at high aH sources of dislocations become active and new dislocations are generated 45. For the ternary alloys Fe-V-C and Fe-Ti-C with maximum strength ∼ 350 MPa the trap density was increased above aH = 30, and the total H content determined by permeation study and by hot extraction were about equal. The formation of dislocations and traps will need higher hydrogen activities, aH > 70 for fine-grain steels with higher strength. For lower values of aH the total H content can be satisfactorily calculated from the permeation data. Such a calculation shows that at aH = 1 the hydrogen in deep traps prevails, whereas at the higher values of aH > 70 the amount of diffusible H becomes higher for all the investigated steels. 3.4 Fracture behaviour in the presence of hydrogen The effect of hydrogen on the fracture of the investigated steels (in table 3) was observed in constant340

extension-rate tests (CERT), conducted in: 1 M H2SO4 + 10-3M As2O3, at charging currents -iC = 0.1 to 8 mA/cm2. The hydrogen activities could be established between aH = 50 and 1500, as was derived from the dependence of aH on iC which was determined before by permeation studies. As a measure of the effect of H on the mechanical properties of steels, the decrease of the work needed for fracturing the steels was measured, i.e. the relative work of fracture Wf = (W/Rm)(W0/Rm0) = Wf / Wf0

(25)

where Rm is the maximum tensile strength, Rm0 without hydrogen. At first a study was conducted on the effect of extension rate µ in tests on specimens which were not precharged with H. The processes leading to the hydrogen effects, H diffusion and local accumulation, H-induced crack formation and crack growth need some time, therefore the extension rate µ must be low enough to realize the hydrogen effects. The results indicate that a rate lower than 10-5/sec should be used to obtain reliable data for Wf, see Figure 7a. Another important result was obtained in tests with different steels at various values of aH, as established by varying the charging current iC. The work of fracture Wf decreases with increasing aH, but at values aH > 500, Wf remains largely constant. Small differences in the hydrogen susceptibility of the steels can be seen from the data in Figure 7b. In addition, a series of tests were performed under three different conditions: a) CERT at aH ∼ 500, without precharging; b) CERT at aH ∼ 1000-1500, without precharging; c) CERT in air (aH = 0) after 96 hours precharging at aH ∼ 1200, and 25 h in air at ambient temperature. MATERIALI IN TEHNOLOGIJE 34 (2000) 6


Series (a) and (b) gave no clear differences in Wf, see Figure 8a, which confirms again that the fracture work becomes independent of aH above aH ∼ 500. Certain differences in H susceptibility can be seen for the different steels, depending on the microstructure. Most important is the result of the last series (c), which gave values for Wf only a little smaller than the original value W°f in absence of hydrogen, or even equal to W°f and higher. While the steels are held in air at room temperature all mobile H will desorb, but the hydrogen in deep traps will stay in the steels. So these experiments clearly prove that it is mainly the mobile H and not the H in deep traps which is responsible for the H effects on fracture. A welding simulation enhances the susceptibility to H, as shown by the tests on one of the steels treated in two different ways: 1. original state, 2. 1 sec 1250 °C + ∆T(800-500 °C) = 15 sec, 3. 1 sec 1250 °C + ∆T (800-500 sec) = 15 sec + 2 hours 600 °C. Both welding simulations cause a significant decrease in Wf, even tempering at 600 °C does not improve the behaviour. 4 CONCLUSIONS Knowledge on hydrogen absorption and diffusion in steels is important for understanding defects and cracking caused by hydrogen in steel. Extensive studies have been conducted on iron, binary and ternary iron alloys and steels, using the electrochemical double-cell method, for determing the H permeation, diffusivity, solubility and activity upon charging the specimens from the atmosphere or from an electrolyte. The effects of the alloying elements C, S, P, Mn, Si, Cr, Ni, Sn and Cu on hydrogen permeation and diffusion have been studied on binary Fe alloys. The permeation coefficient Φ° = cH⋅DH is considerably decreased by Si and Cr, Si reduces the solubility cH and Cr reduces the diffusivity DH by trapping. The surface layer composition after corrosion in 1 M H2SO4 was investigated by AES and XPS, Cu, Sn, P and C are enriched after corrosion, Cr is preferentially dissolved. Cu, Sn and Ni inhibit the dissolution and decrease the hydrogen activity during corrosion. S, P and Mn enhance the corrosion attack, and increase the hydrogen activity. The H diffusivities in Fe alloys and steels are strongly decreased at low temperatures < 200 °C by trapping, i.e. the binding of H to traps at an energy level lower than in the normal interstitial sites. Data on trapping were derived according to the theory of McNabb and Foster 33, from non-steady state permeation studies on binary Fe-Me, ternary Fe-Me-C and Fe-Me-N alloys (Me = Ti, V, Zr, Nb, Mo). For all alloys, flat traps can be found with binding energies around -19 kJ/mol, which are filled to a level more-or-less dependent on aH and affect the diffusion of the mobile H. In contrast, the deep traps with binding energies around -57 kJ/mol are already filled at low aH and the hydrogen in deep traps is MATERIALI IN TEHNOLOGIJE 34 (2000) 6

virtually immobile at ambient temperature. There is no indication of a special chemical bonding of H to carbides or nitrides, it can be assumed that the deep traps are sites in dislocation cores, grain boundaries and at interfaces, whereas the flat traps are extended lattices sites near such defects. These considerations are also true for low-alloy steels, as shown for some pipeline steels. These steels were tested for their hydrogen susceptibility in constantextension-rate tests, conducted after charging with hydrogen from an electrolyte. No great differences were observed in the dependence on their microstructure. The work in fracturing at a low extension rate (10-5/sec) decreased with increasing aH but was about constant at aH > 300. After precharging and desorption of the mobile hydrogen at room temperature there was no effect of the hydrogen in deep traps, on the work of fracture. Only the mobile H in an ideal solution or in flat traps is involved in fracture, its concentration increases approximately in proportion to aH. Transgranular cracks occur in highly stressed regions after mobile H has concentrated sufficiently. The immobile H in deep traps has no effect on the fracture behaviour of the steels, thus there is no direct correlation between total hydrogen content and the susceptibility to hydrogen-induced cracking and stress-corrosion cracking. 5 REFERENCES 1

J. D. Fast: Interaction of Metals and Gases, Academic Press 1965, 126-147 2 R. A. Oriani: Acta Met. 18 (1970) 147 3 R. A. Oriani: Ann. Rev. Matr. Sci. 8 (1978) 327 4 E. Riecke: Arch. Eisenhüttenwes. 44 (1973) 647-656 5 E. Riecke: Werkstoffe u. Korrosion 29 (1978) 106-112 6 E. Riecke: 8th Intern. Congress on Metallic Corrosion, Mainz 1981, Vol. 1, 605-610 7 E. Riecke: 3rd Intern. Congress on Hydrogen in Metals, Paris, June 7-11, 1982 8 J. P. Hirth: Metallurg. Trans. 11 A (1980) 861 9 P. Neumann: Stahl u. Eisen 107 (1987) 577 10 H. Gräfen, D. Kuron: Chem.-Ing.-Tech. 59 (1987) 555 11 A. Turnbull (Ed.): ’Hydrogen Transport and Cracking in Metals’, The Inst. of Materials, London 1995 12 M. H. Armbruster: J. Amer. Chem. Soc. 65 (1943) 1043 13 W. Eichenauer, H. Künzig, A. Pebler: Z. Metallkde. 49 (1958) 220 14 H. Schenck, K. W. Lange: Arch. Eisenhüttenwes. 37 (1966) 739 15 J. O’. M. Bockris, A. K. N. Reddy: Modern Electrochemistry 2, Plenum Press, New York 1977 16 H. Kaesche: Die Korrosion der Metalle, Springer, Heidelberg 1979 17 E. G. Dafft, K. Bohnenkamp, H. J. Engell: Corros. Sci. 19 (1979) 591 18 C. Kato, H. J. Grabke, B. Egert, G. Panzner: Corros. Sci. 24 (1984) 591 19 J. F. Newman, L.L. Shreir: Corr. Sci. 9 (1969) 631 20 J. Albrecht, H.E. Bühler, S. Baumgartl: Archiv Eisenhüttenwes. 45 (1974) 561 21 H. E. Bühler, S. Baumgartl, W. Warnecke: Archiv Eisenhüttenwes. 46 (1975) 811 22 H. Inagaki, M. Tanimura, I. Matsushima, T. Nishimura: Tans. ISIJ. 18 (1979) 149

341

H. J. GRABKE, E. RIECKE: ABSORPTION AND DIFFUSION OF HYDROGEN IN STEELS 23

T. Ramchandran, K. Bohnenkamp: Werkst. Korros. 39 (1979) 43 Z. A. Foroulis: Werkst. u. Korros. 31 (1980) 463 25 R. Pöpperling, W. Schwenk: Werkst. u. Korros. 31 (1980) 15 26 E. Riecke, B. Johnen, H. J. Grabke: Werkstoffe u. Korrosion 36 (1985) 435 27 E. Riecke, R. Möller, B. Johnen, H. J. Grabke: Werkstoffe u. Korrosion 36 (1985) 447 28 E. Riecke, B. Johnen, H. J. Grabke: Werkstoffe u. Korrosion 36 (1985) 455 29 W. Beck, J. O’. M. Bockris, J. Mc.Breen, L. Nanis: Proc. Roy. Soc. A 290 (1966) 220 30 Th. Heumann, E. Domke: Int. Meeting on Hydrogen in Metals, Jülich 1972, Vol. II, 492-515 31 W. Racczynski: Phys. Stat. Sol. (a) 48 (1978) K27 32 N. R. Quick, H. H. Johnson: Acta Met. 26 (1978) 903 33 A. McNabb, P. K. Foster: Trans Metallurg. Soc. AIME 227 (1963) 618/27 34 E. Riecke, R. Möller, B. Johnen: Corros. Sci. 27 (1987) 1027 35 E. Riecke, H. Liesegang, H. J. Grabke: Werkstoffe u. Korrosion 38 (1987) 310-316 24

342

36

E. Riecke, H. Liesegang, H. J. Grabke: Werkstoffe u. Korrosion 38 (1987) 310 37 E. Riecke, B. Johnen, H. Liesegang, A. Thoms, B. Reynders, H. J. Grabke: Werkstoffe u. Korrosion 39 (1988) 525 38 E. Riecke, B. Johnen: Werkstoffe u. Korrosion 42 (1991) 528 39 E. Riecke B. Johnen: Werkstoffe u. Korrosion 42 (1991) 626 40 F. Gehrmann, E. Riecke, H. J. Grabke: Effects of the Nitrides and Carbides of V, Nb and Ti on the Diffusion and Dissolution of Hydrogen in Iron, in (11) p. 216-226. see also: Report EUR 16083DE, Europäische Gemeinschaften, Luxemburg 1997 41 M. A. V. Devanathan, Z. Stachurski: Proc. Roy. Soc. A 270 (1962) 90 42 M. A. V. Devanathan, Z. Stachurski: J. Electrochem. Soc. 111 (1964) 619 43 A. J. Kumnick, H. H. Johnson: Acta Met. 28 (1980) 33 44 G. M. Pressouyre, I.M. Bernstein: Metallurg. Trans. 9A (1978) 1571 45 S. D. Kapusta, T. T. Kam, K. E. Heusler: Z. physik. Chemie N.F. 123 (1980) 219 46 E. Riecke, H. J. Grabke: Einfluß der Mikrostruktur von Stählen auf die Wasserstoffabsorption und wasserstoffinduzierte Rißbildung, Report EUR 16080 DE, Europ. Gemeinschaften Luxemburg 1997
